Embed Size (px)
Citation preview
RNA sensing by conventional dendritic cells is centralto the development of lupus nephritisTeja Celhara, Richard Hopkinsb, Susannah I. Thornhilla, Raquel De Magalhaesa, Sun-Hee Hwangc, Hui-Yin Leea,Hiroko Yasugaa,d, Leigh A. Jonesa,b, Jose Cascoc, Bernett Leea, Thomas P. Thambooe, Xin J. Zhouf, Michael Poidingera,John E. Connollyb,g, Edward K. Wakelandc, and Anna-Marie Fairhursta,c,d,1
aSingapore Immunology Network, A*STAR, 138648 Singapore; bInstitute of Molecular and Cell Biology, A*STAR, 138673 Singapore; cDepartment ofImmunology, University of Texas Southwestern Medical Center, Dallas, TX 75390-9093; dSchool of Biological Sciences, Nanyang Technological University,637551 Singapore; eDepartment of Pathology, National University Hospital, 119074 Singapore; fDepartment of Pathology, Baylor University Medical Center,Dallas, TX 75246; and gInstitute of Biomedical Studies, Baylor University, Waco, TX76798
Edited by Ann Marshak-Rothstein, University of Massachusetts Medical School, Worcester, MA, and accepted by the Editorial Board September 25, 2015(received for review April 14, 2015)
Glomerulonephritis is a common and debilitating feature of sys-temic lupus erythematosus (SLE). The precise immune mechanismsthat drive the progression from benign autoimmunity to glomer-ulonephritis are largely unknown. Previous investigations haveshown that a moderate increase of the innate Toll-like receptor 7(TLR7) is sufficient for the development of nephritis. In thesesystems normalization of B-cell TLR7 expression or temporal de-pletion of plasmacytoid dendritic cells (pDCs) slow progression;however, the critical cell that is responsible for driving full immuno-pathology remains unidentified. In this investigation we have shownthat conventional DC expression of TLR7 is essential for severeautoimmunity in the Sle1Tg7 model of SLE. We show that a novelexpanding CD11b+ conventional DC subpopulation dominates theinfiltrating renal inflammatory milieu, localizing to the glomeruli.Moreover, exposure of human myeloid DCs to IFN-α or Flu increasesTLR7 expression, suggesting they may have a role in self-RNA rec-ognition pathways in clinical disease. To our knowledge, this studyis the first to highlight the importance of conventional DC-TLR7expression for kidney pathogenesis in a murine model of SLE.
TLR7 | dendritic cells | SLE | nephritis | autoimmunity
The innate pathogen recognition receptor Toll-like receptor 7(TLR7) has recently been implicated in the development of
autoimmunity. This receptor, along with TLR8 and TLR9, wasoriginally described by Beutler and colleagues (1), with ligand iden-tification and functional characterization by Akira and colleaguesjust over a year later (2). TLR7 recognizes single-stranded RNA(ssRNA), inducing downstream activation of signaling molecules,including Jnk and NF-κB through a myeloid differentiation pri-mary response gene 88 (MyD88)-dependent cascade (3). Thisprocess is central to host defense against invading viruses;however, TLR7 hyperactivity can also drive the initiation andprogression of autoimmunity.Systemic lupus erythematosus (SLE) is a complex autoimmune
disease characterized by the presence of antinuclear autoanti-bodies (ANA) (4). The ANAs form immune complexes with host-derived nuclear material, which accumulate as deposits within thetissues and organs. The ensuing inflammation can lead to organfailure, particularly glomerulonephritis (GN) (5). Prospective studiesby Harley and colleagues (6) demonstrated that ANAs are de-tectable ∼6 years before clinical presentation of SLE, with only asmall percentage of these individuals who have ANAs progressingto pathogenic autoimmunity. These data, together with murinestudies, suggest that defects in multiple pathways contribute to theinitiation and progression of systemic autoimmunity (7).Multiple investigations have shown that TLR7 and the MyD88
signaling pathway are critical for the initiation of autoimmunityand development of self-reactivity, because genetic ablation ofeither TLR7 or MyD88 prevents the development of both ANAsand subsequent immune-pathology (8–10). Furthermore, this sig-naling pathway is specifically required within B cells (11, 12). In
TLR7-sufficient autoimmune prone systems, additional immunealterations, such as lpr (lymphoproliferation), sle3, or yaa (Y-linkedautoimmune-accelerating locus) can lead to severe autoimmu-nity. The yaamurine susceptibility locus is a region comprising 16genes, including TLR7, which translocated from the X chro-mosome to the Y (13, 14). The yaa-associated increase in TLR7expression and function was determined to be critical for thedevelopment of severe disease, independent of the SLE-model(15–17). More recently, we demonstrated that a twofold increasein TLR7 alone on the Sle1 background is sufficient to drivedisease in an almost identical manner to the addition of the yaasusceptibility locus (18). Normalization of B-cell TLR7 did notaffect GN, suggesting that the increase in other cells drives severepathology. Additional studies have suggested that although plas-macytoid dendritic cells (pDCs) contribute to pathology of a yaa-associated model, other cell types are critical for full diseasepathogenesis (19). Determining these critical cell types willclarify the mechanisms of progression for therapeutic targeting.In this study we investigated the contribution of DCs to TLR7-
mediated SLE-associated GN. Previous data has presentedconflicting evidence on their role in lupus models. AlthoughB-cell proliferation and antibody production can be enhanced byDC activation and genetic ablation diminishes disease progres-sion (20, 21), DCs also play a protective role in regulatory T-cell
Significance
Systemic lupus erythematosus is a complex autoimmune dis-ease where genetic and environmental factors play equallyimportant roles. Recent research has identified that upregula-tion of the innate immune Toll-like receptor 7 (TLR7) is funda-mental for the development of severe disease. In our Sle1Tg7model system, we show that increased expression of TLR7 indendritic cells (DCs) is important for the development of kidneydisease. Furthermore, we identify conventional CD11b+ DCsinfiltrating the glomeruli of diseased mice. In a resting state,human myeloid DCs express negligible levels of TLR7. How-ever, exposure to multiple stimuli results in a dramatic increasein expression. Together, these findings suggest a role for my-eloid DC-TLR7 in clinical disease and that TLR7 expression inthis cellular compartment may represent a novel target fortherapeutic investigation.
Author contributions: T.C., J.E.C., E.K.W., and A.-M.F. designed research; T.C., R.H., S.I.T.,R.D.M., S.-H.H., H.-Y.L., H.Y., L.A.J., J.C., and A.-M.F. performed research; J.C., M.P., J.E.C.,and E.K.W. contributed new reagents/analytic tools; T.C., R.H., S.I.T., L.A.J., B.L., T.P.T., X.J.Z.,M.P., J.E.C., and A.-M.F. analyzed data; and T.C., S.I.T., and A.-M.F. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission. A.M.-R. is a guest editor invited by the EditorialBoard.1To whom correspondence should be addressed. Email: [email protected].
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1507052112/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1507052112 PNAS | Published online October 28, 2015 | E6195–E6204
IMMUNOLO
GYAND
INFLAMMATION
PNASPL
US
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
development and their elimination can result in autoimmunity(22, 23). The overall response in each system may depend on thegenetic background or incomplete ablation of DC subsets [dis-cussed by Platt and Randolph (24)].We bred conditional SLE mice overexpressing TLR7 (Sle1Tg7)
with a CD11cCre reporter strain (25) and demonstrated that theincrease in TLR7 within CD11c+ cells was essential for all severeautoimmune traits, including splenomegaly, T- and B-cell activation,and GN. Purification of DCs and other leukocyte subsets confirmednormalization of TLR7 mRNA in CD11b+ conventional (c)DCs andpDCs. We identified and characterized a novel CD11b+ cDC in-filtrating diseased kidneys, which was absent, together with all otherpathological traits, upon restoration of TLR7. Taken together, ourdata suggest that the CD11b+ cDC, rather than the pDC, is playingthe principle role in driving end organ disease. Finally, we de-termined that primary human BDCA1+ DCs or monocyte-derivedDCs express extremely low levels of TLR7 mRNA, but this can berapidly up-regulated on exposure to virus or IFN-α. This work hasuncovered a key role for DCs in one of the most clinically importantphases of SLE, the progression to lupus nephritis.
ResultsOverexpression of TLR7 in DCs Is Required for the Development ofSevere Autoimmunity in SLE Mice. Previous studies have demonstratedthat B cells are not the critical cell type driving TLR7-dependent
kidney disease in the Sle1Tg7 murine model (18). These findings,together with the conflicting data on the role of DCs in autoimmu-nity, led us to examine the contribution of TLR7 in this compart-ment. CD11cCre-mediated normalization of TLR7mRNA expressionin CD11b+ cDCs and pDCs was confirmed using Nanostring tech-nology (Fig. 1A). No significant differences in TLR7 expression wereobserved in splenic B or peritoneal macrophages, whereas TLR7expression in T cells was low or undetectable (Fig. S1A). Addition-ally, the presence of the CD11cCre did not affect either TLR7mRNAexpression levels (Fig. S1B), splenomegaly, T-cell ratios, or T-cellactivation in Sle1 mice (Fig. S1C). Five- to 7.5-mo-old femaleCD11cCreSle1Tg7 (CD11cSle1Tg7) mice, together with Sle1 andSle1Tg7 controls, were examined for the development of autoim-mune disease. Splenomegaly is a characteristic feature of lupus-prone strains and this was prevented in CD11cSle1Tg7 mice(Fig. 1B). Functional assessment of kidney disease showed normallevels of blood urea nitrogen (BUN) in CD11cSle1Tg7 mice, similarto Sle1 controls (Fig. 1C). Proteinuria did not differ between strains(Fig. S1D). Furthermore, there was no evidence of kidney diseaseby microscopy in either Sle1 or CD11cSle1Tg7 mice, in contrast tothe expected nephritis exhibited by the Sle1Tg7 strain (Fig. 1C).Immunofluorescent staining of kidneys from Sle1Tg7 mice revealedbright glomerular complement (C3d) staining, whereas lower levelswere observed in CD11cSle1Tg7 and Sle1 mice (Fig. S1E). Analysisof renal supernatants from Sle1Tg7 mice revealed significantly
Fig. 1. Prevention of severe autoimmune pathology with CD11c-Cre-normalization of TLR7. (A) TLR7 mRNA expression in pDCs and CD11b+ cDCs assessed by Nanostringin 6- to 8-wk-old mice (n = 4–5 or n = 3 per strain, respectively, for cDC and pDC). Five- to 7.5-mo-old Sle1, Sle1Tg7, and CD11cSle1Tg7 mice were analyzed for: (B) splenicsize (n = 12–13 per strain); (C) concentrations of BUN (n = 13–19 per strain), GN, and interstitial renal disease (IN) scored from 0 to 4 and quantification of glomerularcrescent formation (n= 6–9mice per strain). (D) Analysis of chemokine levels in renal supernatants (n= 12–13 per strain) and (E) IFN-α levels in kidney supernatants by ELISA(n = 11–13 per strain). Bars represent mean + SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and ns, not significant (one-way ANOVA or Kruskal–Wallis test).
E6196 | www.pnas.org/cgi/doi/10.1073/pnas.1507052112 Celhar et al.
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
higher levels of chemokines associated with renal infiltration andGN, including Rantes, MCP-1, and IP-10 (Fig. 1D and Table S1)(18, 26, 27). These were normalized in the CD11cSle1Tg7 strain.IFN-α, which is often associated with pDC activation, was similar inall three strains (Fig. 1E). In contrast to the CD11cCre effects, therewas no reduction in disease pathogenesis upon normalization ofTLR7 using the M-lysozymeCre (LyzMCre) reporter strain, showingthat higher levels of TLR7 in macrophages and neutrophils werenot central in driving severe autoimmune traits (28) (Fig. S2).
Overexpression of TLR7 in DCs Is Required for the Production ofSpecific Subtypes of ANA in SLE Mice. We have previously describedan increase in a variety of autoreactive antibodies in the Sle1 strain,which are augmented by the introduction of either yaa or the Tg7-BAC (18). The predominant IgG autoantibodies are reactive tohistone, chromatin, and double-stranded (ds)DNA. Despite thereduction in kidney pathology, there was no change detected in thetotal ANA titer of CD11cSle1Tg7 compared with Sle1Tg7 mice(Fig. 2, Left). However, both anti-dsDNA anti-snRNP levels werenormalized to Sle1 levels in CD11cSle1Tg7 mice (Fig. 2, Center andRight). These data show that although DC-TLR7 overexpression isnot responsible for the amount of autoantibodies produced in thisSLE model, it does affect the extent of recognition of some self-antigens.
Overexpression of TLR7 in DCs Is Required for Splenic AutoimmuneTraits in SLE Mice.Alongside kidney disease and the production ofANAs, both yaa and Tg7-driven SLE models develop spleno-megaly characterized by leukocyte expansion, including myelo-poiesis. Aged Sle1Tg7 mice demonstrated significant increases inabsolute numbers of CD4+ T cells, B220+ B cells, and CD11b+
myeloid cells within the spleen, compared with Sle1 mice(Fig. 3A). Normalization of DC-TLR7 expression prevented thisexpansion, with leukocyte numbers comparable to those of Sle1mice. Furthermore, CD4+ and CD8+ T cells from Sle1Tg7 mousespleens expressed high levels of the activation markers PD-1,ICOS, and CD69, and the T-cell population was characterized byan increase of effector memory cells (CD62LloCD44hi) (29, 30)concomitant with a reduction in the naïve population (Fig. 3 Dand E and Table S2). These phenomena were either reduced inCD11cSle1Tg7 mice or comparable to Sle1 mice (Fig. 3E andTable S2). Additionally, yaa and TLR7-driven murine lupus modelsare associated with an expansion of the T-follicular helper (Tfh) cellpopulation and a marginal zone B-cell defect: both were detectedin the Sle1Tg7 strain and not in CD11cSle1Tg7 mice (Fig. 3 B andC) (13, 14, 18). Similar trends were observed in expanded pop-ulations of plasmablasts, plasma cells and germinal-center B cells(Table S2).Characterization of CD11b+ populations in the spleen using
flow cytometry indicated that the cell numbers of CD11cSle1Tg7polymorphonuclear leukocytes (PMNs), Gr1+ monocytes, andeosinophils were normalized to Sle1 levels (Table S2).
The remaining Gr1−CD11b+ cells were subdivided accordingto CD11c and MHCII expression (Fig. 3G). Higher frequenciesof CD11c+MHCII+ DCs and CD11c+MHCII− cells were presentin Sle1Tg7 mice, in contrast to CD8+ DCs, which were un-changed (Fig. 3 F and H). This expansion of CD11b+c DCs wasprevented with DC-normalization of TLR7 expression (Fig. 3Hand Table S2). The frequency of pDCs was found to be lowerin Sle1Tg7 mice compared with either Sle1 or Sle1Tg7CD11c(Fig. 3F); however, cell counts reveal that there are no differ-ences in the actual cell numbers across the different strains, in-dicating that the pDC are not an expanding population in thesemice (Table S2).
Overexpression of TLR7 in DCs Is Required for the Infiltration of aNovel CD11c+ Subset into Diseased Kidneys of SLE Mice. Renal pa-thology in SLE mice is associated with an infiltration of CD45+
leukocytes, comprising primarily CD3+ T cells, as well as severalmyeloid populations (18). Normalization of TLR7 expressionlevels in CD11c+ DCs prevented this infiltration (Fig. 4 A and B).The frequency of pDCs, which were identified using the gatingstrategy shown, was not found to be higher in diseased mice(Fig. S3 A and B). Analysis of myeloid cell frequencies in thekidney revealed that CD11cSle1Tg7 mice possessed a significantlylower proportion of eosinophils and Gr1− myeloid cells within theleukocyte population, compared with their Sle1Tg7 counterparts(Fig. 4C). To identify the CD11c+ subpopulations present, wecharacterized the CD45+CD19/3−CD11b+Gr1− subsets in termsof CD11c and MHCII expression (Fig. 4D). We identified twoCD11b+CD11c+ populations in Sle1Tg7 mice that were signifi-cantly more abundant than in Sle1 and DC-TLR7 normalizedmice (Fig. 4E). These populations were F4/80− and eitherMHCII− or MHCII+. Immunofluorescent imaging of the glo-merular leukocyte infiltrate confirmed CD11c+F4/80− cells weremarkedly more abundant in Sle1Tg7 mice (Fig. 4F). In contrast, theabundance of the F4/80+ myeloid subset was not affected by DC-TLR7 normalization (Fig. 4E). Taken together, these data indicatethat overexpression of TLR7 in CD11c-expressing DC drives ac-cumulation of two novel myeloid CD11c+F4/80− subsets in thediseased murine kidney.
Gene Expression Profiles Reveal Distinct Properties of Disease-Associated DC Subsets in SLE Mice. It is unclear whether theCD11c+F4/80+ cells present in all strains are either macrophagesor DCs. They were originally described as a kidney residentsentinel DC (31), but more recently they were reclassified asmacrophages based on their morphological and functional features(32, 33). Therefore, to confirm the ontogeny of these and our novelinfiltrating CD11c+F4/80− populations, we purified the threepopulations identified in Fig. 4 and analyzed their transcriptionalprofiles in diseased Sle1Tg7 mice and Sle1 controls. We sortedsplenic CD11b+ cDC from the same mice for validation pur-poses. Nanostring technology was used to accurately measure theabundance of mRNAs of selected genes associated with DC or
Fig. 2. DC-normalization of TLR7 changes ANA specificity. Sera from 5- to 7.5-mo-old female mice were analyzed for ANA levels by ELISA: IgG dsDNA/histone/chromatin autoantibodies (ANAs), IgG anti-dsDNA Abs (anti-dsDNA), or IgG anti-snRNP Abs (anti-snRNP) (n = 13–20 mice per strain) (AU, absorbance units).Bars represent mean + SEM. *P < 0.05, **P < 0.01, ****P < 0.0001; and ns, not significant (one-way ANOVA or Kruskal–Wallis test).
Celhar et al. PNAS | Published online October 28, 2015 | E6197
IMMUNOLO
GYAND
INFLAMMATION
PNASPL
US
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
macrophage ontogeny, as identified previously (34–37). Addi-tionally, we quantified mRNAs of selected inflammatory genesand several that have been associated with the development ofmurine lupus (Table S3). A two-way ANOVA of normalized datashowed that the expressed genes were differentially expressedamong the four populations, and we therefore subjected them toprincipal component analysis (PCA) (Fig. 5A and Table S3).PCA showed that infiltrating kidney F4/80−MHCII+ cells clusteredtogether with splenic CD11b+ cDCs, whereas kidney-residentF4/80+MHCII+ myeloid cells and infiltrating MHCII− cells formedtwo separate clusters (Fig. 5A). Devolution of these clusters dem-onstrated that renal F4/80−MHCII+ cells expressed core DC-ontogeny genes (including Slamf7, Zbtb46, Kmo, and Btla) (34)and Irf4, an important transcription factor in nonlymphoid tissueCD11b+ DCs (35) (Fig. 5B). Nanostring and flow cytometricanalyses of the renal cDCs demonstrated that they expressedCD64 (Fig. S3 C and D), suggesting that they are monocyte-derived (38, 39). This cDC gene signature was absent from theF4/80+MHCII+ cells, which expressed a typical array of tissuemacrophage-related genes such as Cd64, Mafb, MerTK, C/ebp-a,and Cd14 (36) (Fig. 5C). The infiltrating CD11c+MHCII− pop-ulation, which was particularly abundant in the nephritic kidneys,clustered as a separate population and did not express typicalcDC or macrophage signature genes (Fig. 5 A–C). However, theexpression of monocyte/macrophage-related genes Csfr1r, Lyz2,C/ebp-b, Cd14, and Cx3cr1 points toward a monocyte-derived celltype or a mixed population (36, 40). In summary, these findingsindicate that the resident F4/80+ cells are macrophages and theinfiltrating F4/80−MHCII+ cells are CD11b+cDCs.
All myeloid subsets expressed Tlr7, with the expected increaseexhibited by Sle1Tg7 cells compared with Sle1 controls (Fig. 5D).Tlr9 was also detected but levels were unaffected by Sle1Tg7-associated disease (Fig. 5D). Further analysis of renal MHCII+F4/80− DCs showed a significant down-regulation in Ebi3, Slamf9,Irf4, Rbpj, Batf3, and Kmo in diseased Sle1Tg7 mice (Fig. 5E andTable S4). Analysis of renal macrophages indicated higher levelsof several signaling lymphocyte-activating molecule immunor-eceptor family (SLAMF) members in Sle1Tg7 mice: Cd84, Cracc,Cd229, Cd48, and Cd244 (Table S4). Additionally, macrophagesexpressed the highest amounts of Ccl2 (Mcp-1) transcripts, whichcorrelates with the increased soluble levels in Sle1Tg7 kidneys(Fig. 1D and Table S4). These gene-expression changes suggestthat renal macrophages in diseased mice contribute to the dis-ease progression initiated by the DC.
Isolated Sle1Tg7-Derived Renal DCs Are Functionally Active Cells. Toascertain the functional role of the renal cDCs, we isolated theinfiltrating CD11c+MHCII− and MHCII+F4/80− subsets (Fig. 4D and E) and assessed their ability to present antigen to CD4+ Tcells using the ovalbumin/OT II system. Because the candidategenes for the Sle1 region are homotypic SLAMF members, wegenerated Sle1OTII mice for these studies. We determined thatpurified CD11c+MHCII+F4/80− renal DCs stimulated with TLR7ligand (R848) from all three strains induced T-cell proliferation,whereas the R848-stimulated purified CD11c+MHCII− populationswere unable to induce proliferation above levels of unstimulatedcells (Fig. 6 A and B). Dose–response curves to the TLR7 ligand
Fig. 3. DC-normalization of TLR7 prevents yaa-associated splenic autoimmune traits. Sle1, Sle1Tg7, and CD11cSle1Tg7 mice were analyzed for: (A) Splenicleukocyte counts; major splenic subsets using CD4 and CD8 for T cells, B220, and CD19 for B cells and CD11b for the myeloid lineage. (B) Marginal zone (MZ)B-cell subsets were identified through CD21 and CD23 expression on B220+CD138− cells (18). (C) Tfh were identified as CD4+, CXCR5+, and ICOS+. (D) Naïve andeffector memory CD4+ T cells were gated using CD62L and CD44 expression with (E) the frequencies of CD4+ (Left) and CD8+ T cells (Right). (F) Flowcytometric analysis of splenic pDC and CD8+ DC subsets. (G) Myeloid CD11c+ populations (CD19/CD3/Gr-1−CD11b+), were subgated through MHCII and CD11cinto CD11c+MHCII+ (CD11b+ cDCs) and CD11c+MHCII− cells. (H) Quantification of data from G. Data shown is cumulative from three independent agingcohorts, total n = 9–10 mice per strain (A–H). Plots are representative of one mouse per strain (B, D, and G). Bars represent mean + SEM. *P < 0.05, **P < 0.01,***P < 0.001, ****P < 0.0001; and ns, not significant (one-way ANOVA or Kruskal–Wallis test).
E6198 | www.pnas.org/cgi/doi/10.1073/pnas.1507052112 Celhar et al.
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
were not possible because of the low numbers retrieved fromthe kidney.The secreted cytokines and chemokines following overnight
exposure to R848 were measured using Luminex. Isolated renalDCs from Sle1Tg7 mice secreted MIP-1β, IL-10, MIP-1α, TNF-α,and IL-6 under resting conditions, which were increased furtheron R848 stimulation (Fig. 6C). Normalization of DC-TLR7 ex-pression lowered basal secretion for IL-10 and MIP-1β, withsimilar trends being observed for the remaining mediators. Theresponses following R848 appeared to be intermediate to Sle1and Sle1Tg7; however, this was not significant. These midwayfunctional responses by Sle1Tg7CD11c cells suggests that othercell types expressing the TLR7 transgene may still be impactingthe DC activation state before isolation.Although the Sle1Tg7CD11c DCs still possess a capacity for
antigen presentation and cytokine/chemokine production followingR848 stimulation, there are significantly fewer cells infiltrating thekidney compared with Sle1Tg7 mice, with levels similar to Sle1mice(Fig. 6D). Furthermore, the proportion of DCs in the kidneycorrelates with GN and BUN, which are established measures ofkidney disease (Fig. 6 E and F).
Primary Human BDCA1 and BDCA3 DCs Express TLR7 and Up-RegulateIts Transcription in Response to IFN-α. Previous reports provideconflicting data on the expression of TLR7 in human in vitro
monocyte-derived DCs (mono-DCs) (41, 42). Therefore, we setout to determine the levels of TLR7 in the two subsets of mye-loid DCs that have been described in human peripheral blood:BDCA1+ and BDCA3+. DCs were enriched, sorted, then pooledfrom nine donors and the expression of TLR7, -8, and -9 mRNAswas assessed using RNAseq. Both BDCA1+ and BDCA3+ DCsubsets expressed TLR7, but at levels 10- and 20-fold lower,respectively, than pDCs (Fig. 7A). TLR8 was expressed by bothmyeloid DC subsets, which was consistent with an earlier reportcomparing TLR7, -8, and -9 expression in BDCA1+ DCs andpDCs using two-step real-time PCR (43). Interestingly, the in-creased sensitivity of our approach revealed that both BDCA1+
and BDCA3+ DCs also expressed TLR9, albeit at low levels.Previous studies have shown that TLR7, -8, and -9 expressionmay be modulated by type-I IFN, which is a central mediator inthe host response to viruses and has been associated with thedevelopment of clinical SLE (44–46). Therefore, the purified DCsubsets were exposed to IFN-α, heat inactivated (HI) influenzavirus (Flu), or active PR8-Flu. TLR7 mRNA expression signifi-cantly increased above control levels in BDCA1+ DCs followingtreatment with either IFN-α or PR8, with a concomitant decreasein TLR8 expression (Fig. 7B). BDCA3+ DCs showed minimalresponses to either HI or active Flu, but increased TLR7 anddecreased TLR8 and TLR9 expression in response to IFN-α(Fig. 7C). Exposure of pDCs to all three challenges resulted in
Fig. 4. CD11cCre-normalization of TLR7 prevents renal leukocyte infiltration. (A) Five- to 7.5-mo-old female Sle1, Sle1Tg7, and CD11cSle1Tg7 mice were analyzedfor (A) CD45+ renal leukocytes (n = 12–13 mice per strain, four independent aging cohorts). (B) Major leukocyte subtypes infiltrating the kidneys using CD3 foridentification of T cells (n = 6 mice per strain) and CD11b for the total myeloid population (n = 12–13 mice per strain). (C) Analysis of CD11b+ PMNs, eosinophils,inflammatory Gr1+ monocytes, and remaining Gr1− myeloid cells (n = 12–13 mice per strain). (D) Gating strategy identifying CD11b+CD11c+ populations in thekidney based on CD11c, MHCII and F4/80 expression. Pink gate represents CD11c+MHCII− cells, blue gate represents CD11c+MHCII+F4/80− cells and the green gateidentifies MHCII+F4/80+ cells. (E) Frequencies of CD11b+CD11c+ kidney subsets (n = 12–13 mice per strain). (F) Confocal microscopy showing CD11c+F4/80− cellsinfiltrating the inflamed, enlarged glomeruli. Microscopy magnification: 400×. (Scale bars, 50 μm.) Micrographs are representative of three mice per strain. Barsrepresent mean + SEM. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; and ns, not significant (one-way ANOVA or Kruskal–Wallis test).
Celhar et al. PNAS | Published online October 28, 2015 | E6199
IMMUNOLO
GYAND
INFLAMMATION
PNASPL
US
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
significantly increased TLR7 expression compared with controlsand a concomitant decrease in TLR9 following treatment witheither active or inactive Flu (Fig. 7D). We then generated mono-DCs from 10 healthy human donors and stimulated them overnightwith IFN-α. The cultured GM-CSF/IL-4 mono-DCs were thenanalyzed for TLR7 mRNA expression. We found a significantup-regulation in TLR7 mRNA following IFN-α exposure(Fig. 7E). An earlier report has shown that the addition of IFNαto monocytes cultured with autologous serum induces DC matu-ration. In a similar manner, we detected an increased maturationof the mono-DCs, as demonstrated by the up-regulation of CD80,CD86, and MHC II (Fig. 7F) (47).In summary, these data highlight a particular responsiveness of
primary human and in vitro-derived DCs, which express low basallevels of TLR7 that can be dramatically increased following viralexposure or treatment with IFN-α, a cytokine often implicated inthe development of SLE (48).
DiscussionIn these studies we have shown the central role of the CD11b+cDC in initiating a severe autoimmune response in SLE-pronemice. We have shown that a moderate increase in TLR7 ex-pression in DCs creates a new infiltrating population of CD11b+cDCs that are localized to the glomeruli of nephritic kidneys,creating an inflammatory environment. Moreover, we haveestablished that primary human myeloid DCs and in vitro gen-erated mo-DCs express extremely low levels of TLR7, but mayup-regulate expression in response to ex vivo viral or IFN-α ex-posure. These studies show a central role for CD11b+ cDCs inthe transition from benign autoimmunity to severe pathogenesisin the Sle1Tg7 model system, and suggest this as a valid mech-anism in human disease.To investigate the cellular role of TLR7, we used a conditional
low-copy BAC transgenic TLR7, Tg7, murine strain (18). In com-bination with the sle1 lupus susceptibility region, Sle1Tg7 mice de-velop severe lupus nephritis comparable to other yaa-containinglupus strains, including B6.Sle1Yaa, B6.Nba2Yaa, and B6.YaaFcγRIIB−/− (13, 14, 17, 18). Using the Cre-lox system, we normal-ized expression of TLR7 in CD11c+ DCs and completely preventedthe development of nephritis and all other disease traits associatedwith this model, including splenomegaly, myelopoesis, B- andT-cell activation, and Tfh expansion. Although CD11cCre de-letion normalized expression in both CD11b+ cDCs and pDCs,the absence of IFN-α in diseased kidneys, together with a lack ofpDC expansion in Sle1Tg7 mice, suggests that CD11b+ cDCsplay the more dominant role in driving the inflammatory eventsleading to nephritis. Consistent with this hypothesis, temporal
depletion of pDCs in BXSB mice ameliorates, but does noteliminate all aspects of disease (19).We have previously described a multistep pathway to the de-
velopment of severe autoimmunity (7). The first step is charac-terized by a loss of self-tolerance and the presence of ANAs.This stage is a prerequisite for immune complex deposition,tissue inflammation, and subsequent nephritis. Known geneswithin step 1 include the MHC alleles, the sle1 lupus suscepti-bility region, and SLAMF members. Previous data by a numberof groups has shown that the presence of TLR7 is also a re-quirement for ANA production (9, 11, 12, 49, 50). Moreover,MyD88 signaling, specifically within B cells, is necessary for thedevelopment of ANAs and downstream pathology (10, 12).Furthermore, Reizis and colleagues (51) have recently shownthat pDCs can have a critical role in this initial stage because re-duction of E2-2 alters pDC function to a more tolerogenic pheno-type, which eliminates ANAs altogether and therefore downstreamautoimmune pathogenesis. In the second step of disease pro-gression, known modifiers of pathogenesis include sle3, lpr, andTLR expression variations: yaa and Tg7, which have amplifiedTLR7. These immune alterations alone do not confer end-organpathology but, in combination with pathway 1, result in severedisease. In these studies, we have now shown that a moderate up-regulation of TLR7 within CD11b+ cDCs is necessary to drivethe inflammatory events leading to nephritis in the Sle1Tg7murine lupus model. Thus, TLR7 plays two major roles in amultistep model of disease progression: first, in ANA production,and second, in promoting DC-mediated inflammatory events thatlead to pathogenesis. In a transgenic murine system where TLR7is increased by 8- to 32-fold, additional genes for step 1 are notrequired because TLR7 itself leads to the loss of tolerance (15).As expected, disruption of the humoral immune response usingSAP- or CD40L-deficient systems eliminates germinal centerformation, the production of ANAs, and therefore the devel-opment of GN (52). Interestingly, this high-expressing TLR7transgenic also exhibits a large expansion of CD11b+ cDCs (15)that may contribute to kidney pathogenesis. Our floxed BAC Tg7mice do not develop B-cell autoimmunity in the absence of thesle1 region (18). Furthermore, although modest TLR7 up-regu-lation contributes to leukocyte activation and ANA productionin Sle1Tg7 mice, B-cell normalization does not eliminate path-ogenesis (18).Elimination of MyD88 within DCs in the MRLlpr strain had a
lower impact on the prevention of disease compared with thecentral role they appear to play here (10). The presence of theTLR7 and MyD88 is a prerequisite for ANA production in step 1in our model of autoimmune progression, and this has to occur
Fig. 5. RNA profile of purified CD11b+CD11c+ pop-ulations from the kidney. Cells were sorted fromaged Sle1 and Sle1Tg7 mice, mRNA purified andanalyzed using Nanostring. (A) PCA of gene expres-sion by the indicated kidney and splenic populations(n = 4 per strain, age 7–9 mo). Characterization ofCD11b+CD11c+ subpopulations using genes associatedwith DC (B) or macrophage (C) ontogeny. Heat mapsrepresent the average gene expression level (n = 8)normalized from the lowest (0, dark blue) to thehighest expression (1, bright red). Data visualizationwas performed using TIBCO Spotfire. (D and E) Dif-ferences in expression levels of indicated genes be-tween Sle1 (squares, n = 4) and Sle1Tg7 (circles, n = 4)in the indicated cell subsets *P (adjusted) < 0.05 (seeMaterials andMethods for detailed statistical analysis).
E6200 | www.pnas.org/cgi/doi/10.1073/pnas.1507052112 Celhar et al.
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
before progression to step 2 and severe disease (53). Furthermore,in the MRLlpr model, the Fas mutation (lpr) disrupts the normalprocesses of apoptosis in T cells required for tolerance; the im-mune system becomes overwhelmed with activated autoreactiveT cells, and the role of DCs in directing T-cell activation for path-ogenesis is perhaps less significant than in other SLE models.Thus, different murine models have specific dependencies on
their cellular role, depending on the genes driving the diseaseprogression. Stronger perturbations of the immune system maynot be reflective of the human disease, directing the inflamma-tion more toward one molecule or cell type compared with an-other. Milder alterations involve multiple pathways and cell typesand are harder to decipher, but perhaps are more reflective ofclinical disease. Taking these data together, however, we canunderstand the multiple pathways involved in clinical SLE andreconcile with the heterogeneity of the clinical presentation.We identified three major myeloid CD11b+CD11c+ populations
in the diseased murine kidney. MHCII+F4/80+ macrophages werethe most abundant population in nondiseased kidneys. Thesewere originally described as a tissue-resident sentinel DC population;
however, our data strongly support their reclassification as macro-phages based on gene-expression profiling (31, 54, 55). Examinationof diseased kidneys from Sle1Tg7 mice identified the presence ofCD11c+F4/80− cells specifically within the glomeruli of diseasedmice, suggesting that these cells are critical for the developmentof GN. Analysis of ontogeny gene mRNA expression confirmedthat the infiltrating CD11c+F4/80−MHCII+ subpopulation wereCD11b+cDCs. However, CD11c+F4/80−MHCII− cells did notpossess either a typical DC or macrophage gene signature, andprobably represent a mixed myeloid-infiltrating population,contrary to an earlier investigation that classified these cells asDCs (33).In vitro analyses showed that purified renal DCs were efficient
at presenting antigen and secreted a number of chemokinesunder resting conditions, with higher levels following R848 stimu-lation. Some of the functional phenotypes suggest that the re-sponses by CD11cSle1Tg7 were intermediate to Sle1 and Sle1Tg7.Because these cells were isolated from the kidney, they have beenexposed to an environment where other non-DCs, including the
Fig. 6. Kidney CD11c+MHCII+F4/80− DCs are functional and their numbers correlate with kidney disease. (A) Indicated kidney cell subsets were sorted from5- to 7.5-mo-old mice, stimulated overnight with R848 (or media), pulsed with 50 μg/mL ovalbumin and cocultured with CFSE-stained Sle1.OT-II splenocytes for5 d. The percentages of CD4+TCRVα2+ T-cell proliferation were assessed with flow cytometry by CFSE dilution. (B) Cumulative data from A. Each circle rep-resents kidney cells sorted from an individual mouse (n = 3–4), except for Sle1, where CD11c+MHCII+F4/80− cells were isolated and pooled from two mice each(total of four mice). A paired t test was used to assess statistical significance. (C) Supernatants from R848 (or media) stimulated CD11c+MHCII+F4/80− DCs wereanalyzed by Luminex. (D) Numbers of sorted cells retrieved from kidneys of each strain (n = 5 for Sle1Tg7 and CD11cSle1Tg7, for Sle1 7 mice were pooled intothree groups). Bars represent mean + SEM. *P < 0.05; ns, not significant (one-way ANOVA). Graphs showing the correlation between GN score (E) or BUN (F)and proportion of CD11c+MHCII+F4/80− DCs. The P value and r were computed by nonparametric Spearman correlation.
Celhar et al. PNAS | Published online October 28, 2015 | E6201
IMMUNOLO
GYAND
INFLAMMATION
PNASPL
US
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
kidney cells, may influence DC activation, maturation, and thesubsequent response to inflammatory stimuli.Further characterization of the CD11b+CD11c+MHCII+ DCs
by flow cytometry revealed the presence of surface CD64, sug-gesting that these are monocyte-derived rather than arising di-rectly from a bone marrow precursor (Fig. S3C) (38, 39). Anearlier investigation by Carlin et al. (56) previously demonstratedthat TLR7 enhances the migration of monocytes into the kidney.We propose a mechanism whereby the damage to the kidneycaused by the immune complex deposition results in chemokinesbeing produced by epithelial and endothelial cells. These che-mokines recruit monocytes into the tissue, where the inflammatorymilieu directs differentiation toward a DC phenotype. We haveshown that these Sle1Tg7 DCs actively secrete more inflammatorymediators compared with their Sle1 and Sle1Tg7CD11c counter-parts and their increased presence correlates with establishedmethods of assessing kidney damage (GN and BUN levels), in-dicating their primary role in disease progression.Although the underlying molecular mechanisms driving pa-
thology mediated by DC TLR7 expression are a subject for fu-ture study, several additional changes that we have found in thegene-expression profile of renal DCs in Sle1Tg7 mice may pro-vide some clues: the decreased expression of several myeloiddevelopmental genes may suggest a difference in maturation oractivation of the cells (34, 35, 37). Furthermore, there is a lowerexpression level of Ebi3 (Epstein–Barr virus-induced gene 3),which encodes the EBI3 subunit of IL-27 and IL-35. Deficiencyin the EBI3 subunit has been associated with the development oflupus nephritis and consistent with these findings, an increasewas protective (57, 58).Existing data on the role of TLR7 in human myeloid DCs
suggests a minor role in inflammation: they express extremelylow levels of TLR7 and respond to Flu challenge primarily throughTLR8 (recognizing ssRNA) and cytoplasmic receptors, such asRIG-I (recognizing both ssRNA and dsRNA) (59). However, wefound that primary human BDCA1+ DCs have the capacity tomarkedly up-regulate TLR7 expression in response to ex vivoviral exposure. In support of these findings, monocyte-derivedDCs showed a remarkable capacity to up-regulate TLR7 fol-lowing overnight exposure to IFN-α, a cytokine known to beimportant in clinical SLE (48). Therefore, under these conditions,both TLR7 and TLR8 may act as RNA sensors, each contributingto the inflammatory responses by these cells.In summary, our results demonstrate that up-regulation of
TLR7 within the CD11c+ DC population is required for thedevelopment of severe autoimmunity in Sle1Tg7mice. Restorationof TLR7 expression to normal levels in CD11c+ cells completelyprevented the development of the major autoimmune traits as-
sociated with the model, including splenomegaly with myelopoesisand GN. Furthermore, we have shown that human DCs expresslow levels of TLR7 and have the capacity to up-regulate ex-pression in response to viral stimuli. Moreover, mono-DCs derivedfrom healthy donors up-regulate TLR7 expression in response toIFN-α. These data support our original multistep hypothesis of SLEprogression, where loss of tolerance to self (susceptibility sle1 locus)combined with additional immune alteration (moderate TLR7 up-regulation) culminates in severe kidney disease. Taken together,these findings also suggest that targeting myeloid DC-TLR7 ex-pression may be a productive avenue of research leading to noveltherapeutic options for human SLE.
Materials and MethodsMice.Mice were bred at either the University of Texas Southwestern MedicalCenter or in the Biomedical Resource Centre, Singapore. Breeding pairs forC57BL/6J (B6), CD11cCre, and M-LysozymeCre mice were originally obtainedfrom the Jackson Laboratory (25, 28). The derivation of Sle1 and the gen-eration of the conditional BAC Tg7 have been previously described (18, 60).The sle1 region is flanked by D1Mit17 and D1Mit202 and is NZM2410-derived. For pathological and immunological analysis, female mice wereaged in three to four separate cohorts of ≥3 mice per strain for 5–7.5 mo.Additional cohorts were set up for purification of leukocyte subsets from6- to 8-wk-old mice.
Study Approval. The care and use of laboratory animals conformed to theNational Institutes of Health guidelines and all experimental procedures wereconducted according to an Institutional Animal Care and Use Committee-approved animal protocol. Written informed consent was received fromhealthy donors before inclusion in the study. Ethical approval was obtainedfrom the National University of Singapore (NUS) Institutional Review Board(NUS-IRB 10-250) and Health Sciences Authority HAS-IRB-201306-5 (Singapore).
Assessment of Renal Disease. BUN was assessed using the QuantiChrom UreaAssay Kit (BioAssay Systems). To detect GN, kidneyswere fixed in formalin andembedded in paraffin for blinded analysis by an independent pathologistaccording to World Health Organization specifications, as previously de-scribed (16).
Kidney Supernatant Analysis. Kidneys supernatants were prepared as de-scribed previously (18). Briefly, whole kidneys were minced and resuspendedinto 0.5 mL PBS. Cells were spun down and the supernatants were stored at−80 °C. For IFN-α analysis, supernatants were assayed using a Mouse IFN-alpha Platinum ELISA (Affymetrix) as per the manufacturer’s instructions.The presence of cytokines and chemokines was determined by Luminexmultiplexing technology using the MCYTOMAG-70K MILLIPLEX MAP MouseCytokine/Chemokine Magnetic Bead Panel Immunology Multiplex Assays(Merck Millipore) as per the manufacturer’s instructions.
Fig. 7. Human cDC subsets express low levels ofTLR7 and can up-regulate its expression. Expressionof TLR7, -8, and -9 mRNA assessed by RNAseq in in-dicated human DC subsets under: (A) basal condi-tions and following in vitro stimulation with Live(PR8) or HI Flu or IFN-α compared with a mediacontrol, “C”, in BDCA1+ DCs (B), BDCA3+ DCs (C),and pDCs (D) (n = 3 from nine pooled individuals).(E) GM/IL-4-mono-DCs were stimulated with IFN-α orleft untreated (media). TLR7 expression was mea-sured using RT-PCR and normalized to β-actinexpression and the untreated (media) for eachdonor (n = 10). (F) Flow cytometric analysis of in-dicated surface markers (n = 9). Bars represent mean+ SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and****P < 0.0001 (two-way ANOVA followed byDunnett’s multiple comparisons test). FPKM, frag-ments per kilobase of exon per million mappedreads. (E and F) P values were computed by a Wil-coxon matched-pairs signed rank test, **P < 0.01.
E6202 | www.pnas.org/cgi/doi/10.1073/pnas.1507052112 Celhar et al.
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
Serology. Serum auto-antibodies were measured using ELISAs detecting anti-histone and anti-dsDNA, anti-U1snRNP, or anti-dsDNA IgG, as describedpreviously (16).
Imaging. Kidneys sections from frozen OCT embedded tissues were fixed with1:1 acetone/methanol and stained with antinephrin (AF488), anti-CD11c (PE),and anti-F4/80 (AF647) antibodies. Images were captured with a FV-100confocal system comprising an Olympus IX81 microscope and FV1000 scanhead and processed with Olympus FV10-ASW Fluoview v2.0b.
Flow Cytometry and Cell Sorting of Murine Cells. A single-cell suspension wasprepared from dissected kidneys, as described previously (16). Splenocytes orkidney cells were resuspended in PBS with 1% (vol/vol) FCS and labeled witha combination of up to 12 directly conjugated antibodies. Red blood cell(RBC) lysis was achieved using BD FACS Lysing solution as per manufacturer’sinstructions. To identify murine splenic DCs, we gated on CD45 expressionthen analyzed B220 and CD11b expression, excluding cells expressing CD19and CD3 (17A2) before taking into account subtype-specific markers: pDCswere defined as B220+CD11b−CD19−CD11cloSiglecH+Gr1+; CD8+ DCs wereB220−CD11b−CD8+CD19/3−CD11c+MHCII+; and total myeloid/conventionalDCs were CD11b+B220−Gr1−Side-scatterloCD11c+MHCII+. Renal myeloid sub-sets were CD45+CD11b+CD11c+ and CD3/19−Gr1−. Further gating identifiedMHCII− and MHCII+ populations and the MHCII+ population was subdividedbased on F4/80 expression. Acquisition and analysis was completed using a BDLSR II with Flowjo 7.6 for Windows (Treestar). For sorting, RBC lysis wasachieved with ACK lysing buffer (Lonza) and live DAPI− cells were sorted usinga BD FACSAria system. Antibodies were purchased either from BD Biosciences,eBioscience, BioLegend, or Life Technologies.
Nanostring Gene-Expression Analysis. The nCounter Analysis System byNanostring allows absolute quantification of very small amounts of mRNAwithout an amplification step, which can introduce variability and bias (61).Following fluorescent labeling procedures described above, murine kidneycells or splenocytes were sorted on a BD FACSAria in the SIgN Flow Cytom-etry Core Facility according to the gating strategy described above andshown in the Results. Fifteen-thousand cells in 5 μL of RLT buffer (Qiagen)with 1% (vol/vol) 2-mercaptoethanol were used for the Gene ExpressionAssay using the Nanostring nCounter system, as per the manufacturer’srecommended protocol. Probes were designed by Nanostring (standard as-say) with the gene-expression list shown in Table S3. The housekeepinggenes B2m, Cltc, and Tbp were used to normalize for RNA loading differ-ences. A two-way ANOVA with post hoc Tukey’s test was used for each genewith the cell type and the Tg7 transgene status as the factors together withtheir interaction term. Student’s t test with Welch correction was used todetermine significant differences in the expression of each gene betweentransgenic versus nontransgenic within each cell type. Multiple testing cor-rection was done using the method of Benjamini and Hochberg with P <0.05 postcorrection being considered significant. All of the statistical analysiswas conducted using the R statistical language version 2.15.2. Data visuali-zation was performed using TIBCO Spotfire.
Human DC Isolation and Stimulation. Peripheral blood mononuclear cells(PBMCs) were obtained from blood of healthy donors by Ficoll-Hypaquecentrifugation (Amersham Pharmacia Biotech). Primary human DCs wereenriched from PBMCs, using a Dynabead Human DC Kit (Life Tech-nologies). Enriched DCs were further purified by FACS: BDCA1 subset(Lin1−HLADR+CD11c+BDCA1+) BDCA3+ (Lin1−HLADR+CD11c+BDCA3+),and pDCs (Lin1−HLADR+CD123+CD11c−). Cells were then seeded in a 96-wellplate to give a minimum cell number of 1 × 104 cells per well. DC subsetsfrom individual donors were stimulated separately with either IFN-a2b(1000 U; Miltenyi Biotec), live PR8 (0.512:1 HAU; Charles River), heat-inactivatedPR8 (0.512:1 HAU) or were not stimulated, for 8 h. Cells were subsequentlylysed in mirVana buffer and stored at –80 °C until RNA isolation and RNAseqanalysis.
RNA Isolation and RNA-Seq. Human DCs were isolated from nine donors asdescribed above. Total RNA was isolated from the DC pellets using amirVanamiRNA Isolation kit, according to the manufacture’s guidelines, and ana-lyzed on Agilent Bioanalzyer for quality assessment. Samples were pooledinto three groups; cDNA libraries were prepared using 24–30 ng of total RNAand 2 μL of a 3:5,000 dilution of External RNA Controls Consortium (ERCC)RNA Spike in Controls (Ambion). The fragmented mRNA samples weresubjected to cDNA synthesis using Illumina TruSeq RNA sample preparationkit version 1 (low-throughput protocol) according to the manufacturer’sprotocol, with the following modifications: (i) use of 14 PCR cycles, and
(ii) two additional rounds of Agencourt Ampure XP SPRI beads (BeckmanCourter) to remove >600 bp double-stranded cDNA.
RNAseq Analysis. RNAseq fastq paired end data were first quality checkedwith fastqc. (www.bioinformatics.babraham.ac.uk/projects/fastqc/). Datawere then mapped to an ERCC spikein database with Bowtie using defaultparameters except max-insert-size set to 500 (www.thermofisher.com/order/catalog/product/4456740) (62). Picard tools (picard.sourceforge.net/) wereused to determine the mean mate distance and SD for each sample. The datawere mapped to HG19 with TopHat2 (tophat.cbcb.umd.edu (63) using de-fault parameters, except for the mate information, Gencode V16 annotationas a guide (www.gencodegenes.org/), with no novel junctions. A customscript based on the R subreads package was then used to generate rawcounts per exon using the gencode V16 annotation, accumulated at thegene level. Counts were normalized by fragments per kilobase of exon permillion mapped reads. edgeR (64) was used to determine differential geneexpression on a pairwise basis. All of this methodology was written withinthe Pipeline Pilot frame work (accelrys.com). Statistical significance was de-termined using a two-way ANOVA followed by Dunnett’s multiple com-parisons test in GraphPad Prism 6.05.
Antigen Presentation. Kidney CD11c+MHCII− and CD11c+MHCII+F4/80− cellswere sorted as described, resuspended in complete media, and seeded inround-bottom 96-well plates at 104cells per well. Where indicated, cellssorted from multiple mice were pooled to achieve sufficient cell numbers.Cells were stimulated in overnight cultures with 1 μg/mL R848 (InvivoGen)or left untreated, and then incubated with 50 μg/mL ovalbumin (Sigma-Aldrich). After 4 h, the media was replaced with a suspension of CFSE-stained splenocytes from Sle1.transgenic OT-II T-cell receptor mice at2 × 105 cells per well. After 5 d of incubation, the proliferation of T cellswas assessed by CFSE dilution using flow cytometry. Supernatants fromovernight stimulations (media and R848) were collected and immediatelyfrozen at −80°. The presence of cytokines and chemokines was assessed byLuminex as described for kidney supernatants.
Monocyte-Derived Dendritic Cell Isolation and Stimulation. PBMCs wereobtained from healthy donor blood as described above and CD14+monocyteswere purified using a CD14-microbead kit (Miltenyi Biotec). Cells were cul-tured in a six-well plate at 2 × 105 cells per well, 2 mL per well in RPMIcomplete medium (RPMI CM; RPMI-1640 containing 10% (vol/vol) heat-inactivated FCS, 10 mM Hepes, 2 mM L-glutamine, 100 IU/mL penicillin, and100 μg/mL streptomycin) with GM-CSF (1,000 U/mL) and IL-4 (1,000 U/mL).Fresh medium and cytokines were added to the culture on day 4. Mono-DCswere harvested and seeded on day 6 in a 96-well U-bottom plate at 1 × 105
cells per well, before stimulation with IFN-a2b (1,000 U), or left unstimu-lated, for 16 h. Cells were then collected by centrifugation and an aliquotwas stained for flow cytometric analysis with antibodies directed to CD80,CD86, and HLADR (BD Pharmingen). A near IR Live Dead Dye (Life Tech-nologies) was used to gate out dead cells. The rest of the cells were resus-pended in RLT buffer (Qiagen), homogenized by syringing and immediatelyfrozen on dry ice. Cell lysates were stored at −80 °C until RNA isolation.
Quantitative Real-Time PCR. RNA was extracted using the RNeasy Micro Kit(Qiagen) and concentrations were measured using Nanodrop 2000 spec-trophotometer (Thermo Scientific). RNA was reverse-transcribed and RT-PCRwas performed using the Power SYBR Green PCR master mix (Applied Bio-systems) on the 7900H fast real-time PCR system (Applied Biosystems). Primersfor human TLR7 were from Invivogen (catalog #rtp-htlr7) and primers forβ-actin were as described previously (65). The results were analyzed usingSDS 2.4, Microsoft Excel for Windows, and GraphPad Prism v6.05.
Statistics. Nanostring and RefSeq data were analyzed as described. Data areexpressed as the arithmetic mean + SEM andwere statistically analyzed usingGraphPad Prism 6.05. When comparing two groups, multiple t tests wereused, corrected for multiple comparisons with the Holm–Sidak method.Where indicated, a paired t test or Wilcoxon matched-pairs signed rank testwas used to compute P values for paired data. When three groups werecompared, parametric data were assessed by ordinary one-way ANOVA withpost hoc Bonferroni’s multiple comparison test and nonparametric datawere assessed with the Kruskal–Wallis test with post hoc Dunn’s multiplecomparisons test. Correlation was assessed by Spearman’s rank correlationtest. Where there is no notation of significance on a figure, there was nosignificant difference found.
Celhar et al. PNAS | Published online October 28, 2015 | E6203
IMMUNOLO
GYAND
INFLAMMATION
PNASPL
US
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
ACKNOWLEDGMENTS. We thank Dr. Lucy Robinson of Insight EditingLondon for critical review and assistance in manuscript preparation; FlorentGinhoux for discussions on myeloid ontogeny; and the Flow Cytometry andFunctional Genomic cores at the Singapore Immunology Network, A*STAR,Singapore, and the flow cytometry core at the University of Texas South-
western Medical Center, Dallas, TX. This work was supported by corefunding from the Singapore Immunology Network and Institute of Celland Molecular Biology at A*STAR, Singapore (to A.-M.F. and J.C.), andgrants from the National Institutes of Health and the Alliance for LupusResearch (to E.K.W.).
1. Du X, Poltorak A, Wei Y, Beutler B (2000) Three novel mammalian toll-like receptors:Gene structure, expression, and evolution. Eur Cytokine Netw 11(3):362–371.
2. Hemmi H, et al. (2002) Small anti-viral compounds activate immune cells via the TLR7MyD88-dependent signaling pathway. Nat Immunol 3(2):196–200.
3. Barton GM, Medzhitov R (2003) Toll-like receptor signaling pathways. Science300(5625):1524–1525.
4. Tan EM, et al. (1982) The 1982 revised criteria for the classification of systemic lupuserythematosus. Arthritis Rheum 25(11):1271–1277.
5. Maroz N, Segal MS (2013) Lupus nephritis and end-stage kidney disease. Am J Med Sci346(4):319–323.
6. Arbuckle MR, et al. (2003) Development of autoantibodies before the clinical onset ofsystemic lupus erythematosus. N Engl J Med 349(16):1526–1533.
7. Fairhurst AM, Wandstrat AE, Wakeland EK (2006) Systemic lupus erythematosus:Multiple immunological phenotypes in a complex genetic disease. Adv Immunol 92:1–69.
8. Nickerson KM, et al. (2010) TLR9 regulates TLR7- and MyD88-dependent auto-antibody production and disease in a murine model of lupus. J Immunol 184(4):1840–1848.
9. Santiago-Raber ML, et al. (2010) Critical role of TLR7 in the acceleration of systemiclupus erythematosus in TLR9-deficient mice. J Autoimmun 34(4):339–348.
10. Teichmann LL, Schenten D, Medzhitov R, Kashgarian M, Shlomchik MJ (2013) Signalsvia the adaptor MyD88 in B cells and DCs make distinct and synergistic contributionsto immune activation and tissue damage in lupus. Immunity 38(3):528–540.
11. Berland R, et al. (2006) Toll-like receptor 7-dependent loss of B cell tolerance inpathogenic autoantibody knockin mice. Immunity 25(3):429–440.
12. Soni C, et al. (2014) B cell-intrinsic TLR7 signaling is essential for the development ofspontaneous germinal centers. J Immunol 193(9):4400–4414.
13. Pisitkun P, et al. (2006) Autoreactive B cell responses to RNA-related antigens due toTLR7 gene duplication. Science 312(5780):1669–1672.
14. Subramanian S, et al. (2006) A Tlr7 translocation accelerates systemic autoimmunity inmurine lupus. Proc Natl Acad Sci USA 103(26):9970–9975.
15. Deane JA, et al. (2007) Control of toll-like receptor 7 expression is essential to restrictautoimmunity and dendritic cell proliferation. Immunity 27(5):801–810.
16. Fairhurst AM, et al. (2008) Yaa autoimmune phenotypes are conferred by over-expression of TLR7. Eur J Immunol 38(7):1971–1978.
17. Santiago-Raber ML, et al. (2008) Evidence for genes in addition to Tlr7 in the Yiaatranslocation linked with acceleration of systemic lupus erythematosus. J Immunol181(2):1556–1562.
18. Hwang SH, et al. (2012) B cell TLR7 expression drives anti-RNA autoantibody pro-duction and exacerbates disease in systemic lupus erythematosus-prone mice.J Immunol 189(12):5786–5796.
19. Rowland SL, et al. (2014) Early, transient depletion of plasmacytoid dendritic cellsameliorates autoimmunity in a lupus model. J Exp Med 211(10):1977–1991.
20. Sang A, et al. (2014) Dysregulated cytokine production by dendritic cells modulates Bcell responses in the NZM2410 mouse model of lupus. PLoS One 9(8):e102151.
21. Teichmann LL, et al. (2010) Dendritic cells in lupus are not required for activation of Tand B cells but promote their expansion, resulting in tissue damage. Immunity 33(6):967–978.
22. Birnberg T, et al. (2008) Lack of conventional dendritic cells is compatible with normaldevelopment and T cell homeostasis, but causes myeloid proliferative syndrome.Immunity 29(6):986–997.
23. Ohnmacht C, et al. (2009) Constitutive ablation of dendritic cells breaks self-toleranceof CD4 T cells and results in spontaneous fatal autoimmunity. J Exp Med 206(3):549–559.
24. Platt AM, Randolph GJ (2010) Does deleting dendritic cells delete autoimmunity?Immunity 33(6):840–842.
25. Caton ML, Smith-Raska MR, Reizis B (2007) Notch-RBP-J signaling controls the ho-meostasis of CD8− dendritic cells in the spleen. J Exp Med 204(7):1653–1664.
26. Haberstroh U, et al. (2002) Expression of the chemokines MCP-1/CCL2 and RANTES/CCL5 is differentially regulated by infiltrating inflammatory cells. Kidney Int 62(4):1264–1276.
27. Romagnani P, et al. (2002) IP-10 and Mig production by glomerular cells in humanproliferative glomerulonephritis and regulation by nitric oxide. J Am Soc Nephrol13(1):53–64.
28. Clausen BE, Burkhardt C, Reith W, Renkawitz R, Förster I (1999) Conditional genetargeting in macrophages and granulocytes using LysMcre mice. Transgenic Res 8(4):265–277.
29. Sallusto F, Lenig D, Förster R, Lipp M, Lanzavecchia A (1999) Two subsets of memoryT lymphocytes with distinct homing potentials and effector functions. Nature401(6754):708–712.
30. Wherry EJ, et al. (2003) Lineage relationship and protective immunity of memory CD8T cell subsets. Nat Immunol 4(3):225–234.
31. Soos TJ, et al. (2006) CX3CR1+ interstitial dendritic cells form a contiguous networkthroughout the entire kidney. Kidney Int 70(3):591–596.
32. Lionakis MS, et al. (2013) CX3CR1-dependent renal macrophage survival promotesCandida control and host survival. J Clin Invest 123(12):5035–5051.
33. Sahu R, Bethunaickan R, Singh S, Davidson A (2014) Structure and function of renalmacrophages and dendritic cells from lupus-prone mice. Arthritis Rheumatol 66(6):1596–1607.
34. Miller JC, et al.; Immunological Genome Consortium (2012) Deciphering the tran-scriptional network of the dendritic cell lineage. Nat Immunol 13(9):888–899.
35. Schlitzer A, et al. (2013) IRF4 transcription factor-dependent CD11b+ dendritic cells inhuman and mouse control mucosal IL-17 cytokine responses. Immunity 38(5):970–983.
36. Gautier EL, et al.; Immunological Genome Consortium (2012) Gene-expression profilesand transcriptional regulatory pathways that underlie the identity and diversity ofmouse tissue macrophages. Nat Immunol 13(11):1118–1128.
37. Guilliams M, et al. (2014) Dendritic cells, monocytes and macrophages: a unifiednomenclature based on ontogeny. Nat Rev Immunol 14(8):571–578.
38. Langlet C, et al. (2012) CD64 expression distinguishes monocyte-derived and con-ventional dendritic cells and reveals their distinct role during intramuscular immu-nization. J Immunol 188(4):1751–1760.
39. Plantinga M, et al. (2013) Conventional and monocyte-derived CD11b(+) dendriticcells initiate and maintain T helper 2 cell-mediated immunity to house dust mite al-lergen. Immunity 38(2):322–335.
40. Geissmann F, Jung S, Littman DR (2003) Blood monocytes consist of two principalsubsets with distinct migratory properties. Immunity 19(1):71–82.
41. Larangé A, Antonios D, Pallardy M, Kerdine-Römer S (2009) TLR7 and TLR8 agoniststrigger different signaling pathways for human dendritic cell maturation. J LeukocBiol 85(4):673–683.
42. Rissoan MC, et al. (1999) Reciprocal control of T helper cell and dendritic cell differ-entiation. Science 283(5405):1183–1186.
43. Guiducci C, et al. (2013) RNA recognition by human TLR8 can lead to autoimmuneinflammation. J Exp Med 210(13):2903–2919.
44. Derkow K, et al. (2013) Multiple sclerosis: Modulation of Toll-like receptor (TLR) ex-pression by interferon-β includes upregulation of TLR7 in plasmacytoid dendritic cells.PLoS One 8(8):e70626.
45. Bekeredjian-Ding IB, et al. (2005) Plasmacytoid dendritic cells control TLR7 sensitivityof naive B cells via type I IFN. J Immunol 174(7):4043–4050.
46. Rönnblom L, Pascual V (2008) The innate immune system in SLE: Type I interferonsand dendritic cells. Lupus 17(5):394–399.
47. Blanco P, Palucka AK, Gill M, Pascual V, Banchereau J (2001) Induction of dendritic celldifferentiation by IFN-alpha in systemic lupus erythematosus. Science 294(5546):1540–1543.
48. Pascual V, Farkas L, Banchereau J (2006) Systemic lupus erythematosus: All roads leadto type I interferons. Curr Opin Immunol 18(6):676–682.
49. Christensen SR, et al. (2006) Toll-like receptor 7 and TLR9 dictate autoantibodyspecificity and have opposing inflammatory and regulatory roles in a murine modelof lupus. Immunity 25(3):417–428.
50. Groom JR, et al. (2007) BAFF and MyD88 signals promote a lupuslike disease in-dependent of T cells. J Exp Med 204(8):1959–1971.
51. Sisirak V, et al. (2014) Genetic evidence for the role of plasmacytoid dendritic cells insystemic lupus erythematosus. J Exp Med 211(10):1969–1976.
52. Walsh ER, et al. (2012) Dual signaling by innate and adaptive immune receptors isrequired for TLR7-induced B-cell-mediated autoimmunity. Proc Natl Acad Sci USA109(40):16276–16281.
53. Celhar T, Magalhães R, Fairhurst AM (2012) TLR7 and TLR9 in SLE: When sensing selfgoes wrong. Immunol Res 53(1-3):58–77.
54. Krüger T, et al. (2004) Identification and functional characterization of dendritic cellsin the healthy murine kidney and in experimental glomerulonephritis. J Am SocNephrol 15(3):613–621.
55. Taylor PR, et al. (2005) Macrophage receptors and immune recognition. Annu RevImmunol 23:901–944.
56. Carlin LM, et al. (2013) Nr4a1-dependent Ly6C(low) monocytes monitor endothelialcells and orchestrate their disposal. Cell 153(2):362–375.
57. Igawa T, et al. (2009) Deficiency in EBV-induced gene 3 (EBI3) in MRL/lpr mice resultsin pathological alteration of autoimmune glomerulonephritis and sialadenitis. ModRheumatol 19(1):33–41.
58. Shinsuke N, Hiroshi I (2013) Overexpression of Epstein-Barr virus-induced gene 3protein (EBI3) in MRL/lpr mice suppresses their lupus nephritis by activating regula-tory T cells. Autoimmunity 46(7):446–454.
59. Jensen S, Thomsen AR (2012) Sensing of RNA viruses: A review of innate immunereceptors involved in recognizing RNA virus invasion. J Virol 86(6):2900–2910.
60. Morel L, et al. (1997) Functional dissection of systemic lupus erythematosus usingcongenic mouse strains. J Immunol 158(12):6019–6028.
61. Kulkarni MM (2011) Digital multiplexed gene expression analysis using the Nano-String nCounter system. Curr Protoc Mol Biol Chapter 25:Unit25B.
62. Langmead B, Salzberg SL (2012) Fast gapped-read alignment with Bowtie 2. NatMethods 9(4):357–359.
63. Kim D, et al. (2013) TopHat2: Accurate alignment of transcriptomes in the presence ofinsertions, deletions and gene fusions. Genome Biol 14(4):R36.
64. Robinson MD, McCarthy DJ, Smyth GK (2010) edgeR: A Bioconductor package for differ-ential expression analysis of digital gene expression data. Bioinformatics 26(1):139–140.
65. Kokkinopoulos I, Jordan WJ, Ritter MA (2005) Toll-like receptor mRNA expressionpatterns in human dendritic cells and monocytes. Mol Immunol 42(8):957–968.
E6204 | www.pnas.org/cgi/doi/10.1073/pnas.1507052112 Celhar et al.
Dow
nloa
ded
by g
uest
on
Nov
embe
r 2,
202
0
